For consultancies, agencies and professional firms
distill.pm distills your email, chat and documents into a playbook your whole team can consult.
Nobody writes anything.

The partner who resigns on Friday takes nineteen years of judgment with her.
The bid you price tomorrow repeats a mistake from the one you lost last year.
A new hire spends their first year learning how the firm works by asking around.
Somebody built a wiki once. Nobody maintains it and nobody trusts it.
The answers already exist. They are sitting in years of email, chat and documents.
From the playbook
Every lesson looks like this: the rule, the real occasions it came from, and the documents one click away.

Ask
A question typed before a bid, and the answer assembled from the firm's own rules, each cited with its count. When the playbook has nothing on a question, it says so.

The benchmark
A general AI model already makes your firm's call about four times in ten. The other six live only in your archive. We measure the split on your own material, so you know exactly what you're sitting on.
Asked situations from your firm's own history, a general AI model makes a call your experience says is worse about half the time. That gap is what your archive is worth.

What you get
01
Every lesson your firm has learned: the rule, the real occasions it was learned from, each with what happened, how it ended, and the original documents a click away. When someone leaves, their lessons stay.

02
Put a situation to it, get your firm's answer, with the lessons it used cited beside the reply. Working a bid, it also lines up the material worth reusing: the closest past proposals, the case-study slides, the answers you wrote last time, each a click from its source.

03
For the situations that keep coming back: pricing a bid, kicking off, a client threatening to walk. A play holds the lessons in the order they apply. Read it on the way in.

04
On a schedule, distill.pm reads what's new, adds lessons, and enriches old ones with new examples, including the ones that went wrong. When the firm quietly stops following one of its own rules, it notices, and asks you whether that was deliberate. You review only what it is unsure about.

Day one
A new hire opens distill.pm on day one: the house rules the firm has kept for years, then the plays for their role, each backed by real cases. Learning by asking around stops being the plan.

How it works
01
Google Workspace, Microsoft 365, Slack, or upload an archive.
02
The situations where judgment was exercised become lessons, with the source documents a click away.
03
Confirm new lessons, reword the ones that have changed. Minutes, not days.
04
Playbook, chat, API.
We read, we distill, we discard. distill.pm keeps lessons and pointers, not your archive. Your documents stay in your systems.
The identity layer
A decade of mail mentions five different Pauls, three nicknames and a client who changed companies twice. distill.pm builds the who's-who of your firm's history, so every lesson stays traceable to the people and clients it came from. Names are replaced with neutral labels the moment documents are read, kept apart from the lessons, and shown only to the readers you permit. In a private build, they never leave your systems at all.

The API
The whole playbook is available over a read-only API and as an MCP server your own AI agents can plug into. An AI assistant drafting a proposal pulls the applicable lessons and the reusable material the same way a person would, and cites them the same way too.

Built on ourselves first: twelve years of a real consultancy's email, chat and documents.
Years of our own archive
Documents read
Lessons, with receipts
Prefer mail? info@distill.pm
distill.pm · a Fluxus product · info@distill.pm